Author: jukka
Date: Fri Sep 2 10:59:39 2011
New Revision: 1164471
URL: http://svn.apache.org/viewvc?rev=1164471&view=rev
Log:
TIKA-207: MS word doc containing tracked changes produces incorrect text
Patch by Curt Arnold
Modified:
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/WordExtractor.java
Modified:
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/WordExtractor.java
URL:
http://svn.apache.org/viewvc/tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/WordExtractor.java?rev=1164471&r1=1164470&r2=1164471&view=diff
==============================================================================
---
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/WordExtractor.java
(original)
+++
tika/trunk/tika-parsers/src/main/java/org/apache/tika/parser/microsoft/WordExtractor.java
Fri Sep 2 10:59:39 2011
@@ -214,7 +214,7 @@ public class WordExtractor extends Abstr
private void handleCharacterRun(CharacterRun cr, boolean skipStyling,
XHTMLContentHandler xhtml)
throws SAXException {
// Skip trailing newlines
- if(cr.text().equals("\r"))
+ if(!isRendered(cr) || cr.text().equals("\r"))
return;
if(!skipStyling) {
@@ -347,7 +347,7 @@ public class WordExtractor extends Abstr
private void handlePictureCharacterRun(CharacterRun cr, Picture picture,
PicturesSource pictures, XHTMLContentHandler xhtml)
throws SAXException, IOException, TikaException {
- if(picture == null) {
+ if(!isRendered(cr) || picture == null) {
// Oh dear, we've run out...
// Probably caused by multiple \u0008 images referencing
// the same real image
@@ -471,6 +471,17 @@ public class WordExtractor extends Abstr
}
/**
+ * Determines if character run should be included in the extraction.
+ *
+ * @param cr character run.
+ * @return true if character run should be included in extraction.
+ */
+ private boolean isRendered(final CharacterRun cr) {
+ return cr == null || !cr.isMarkedDeleted();
+ }
+
+
+ /**
* Provides access to the pictures both by offset, iteration
* over the un-claimed, and peeking forward
*/
